1. Field of the Invention
This invention relates to an ultra-low force atomic force microscope, and particularly an improvement to the atomic force microscope described in related commonly owned U.S. patent application Ser. No. 08/147,571 and related U.S. Pat. Nos. 5,229,606 and 5,266,801.
2. Discussion of the Background
Atomic Force Microscopes (AFM's) are extremely high resolution surface measuring instruments. Two types of AFM's have been made in the past, the contact mode (repulsive mode) AFM and the non-contact (attractive mode) AFM.
The contact mode AFM is described in detail in U.S. Pat. No. 4,935,634 by Hansma et al, as shown in FIG. 2. This AFM operates by placing a sharp tip attached to a bendable cantilever directly on a surface and then scanning the surface laterally. The bending of the lever in response to surface height variations is monitored by a detection system. Typically, the height of the fixed end of the cantilever relative to the sample is adjusted with feedback to maintain the bending at a predetermined amount during lateral scanning. The adjustment amount versus lateral position creates a map of the surface. The deflection detection system is typically an optical beam system as described by Hansma et al. Using very small microfabricated cantilevers and piezoelectric positioners as lateral and vertical scanners, AFM's can have resolution down to molecular level, and may operate with controllable forces small enough to image biological substances. Since AFM's are relatively simple, inexpensive devices compared to other high resolution techniques and are extremely versatile, they are becoming important tools in a wide variety of research and high technology manufacturing applications. The contact mode AFM, in which the tip is maintained in continuous contact with the sample, is currently the most common type, and accounts for essentially all the AFM's sold commercially to date.
The contact AFM has found many applications. However, for samples that are very soft or interact strongly with the tip, such as photoresist, some polymers, silicon oxides, many biological samples, and others, the contact mode has drawbacks. As pointed out in Hansma et al, the tip may be attracted to the surface by the thin liquid layer on all surfaces in ambient conditions, thus increasing the force with which the tip presses on the surface. The inventors and others have also observed that electrostatic forces may attract the tip to the surface, particularly for some tip-sample combinations such as silicon nitride tips on silicon oxide surfaces. When the tip is scanned laterally under such conditions, the sample experiences both compressive and shearing forces. The lateral shearing forces may make the measurement difficult and for soft samples may damage the sample. Further, a stick-slip motion may cause poor resolution and distorted images. Hansma et al's approach to this problem was to immerse the tip, cantilever, and sample surface in liquid, thus eliminating the surface layer forces, and for a polar liquid, the electrostatic forces. This technique works very well, and has the further advantage that it allows samples that are normally hydrated to be imaged in their natural state. However for many samples and applications, immersion in liquid is not of much use. Operating in liquid requires a fluid cell and increases the complexity of using the AFM, and for industrial samples such as photoresist and silicon wafers, immersion is simply not practical.
The non-contact AFM, developed by Martin et al, J. Applied Physics, 61(10), 15 May, 1987, profiles the surface in a different fashion than the contact AFM. In the non-contact AFM, the tip is scanned above the surface, and the very weak Van der Waals attractive forces between the tip and sample are sensed. Typically in non-contact AFM's, the cantilever is vibrated at a small amplitude and brought near to the surface such that the force gradient due to interaction between the tip and surface modifies the spring constant of the lever and shifts its natural resonant frequency. The shift in resonance will change the cantilever's response to the vibration source in a detectable fashion. Thus the amount of change may be used to track the surface typically by adjusting the probe surface separation during lateral scanning to maintain a predetermined shift from resonance. This AC technique provides greater sensitivity than simply monitoring the DC cantilever deflection in the presence of the attractive Van der Waals force due to the weak interaction between the tip and surface. The frequency shift may be measured directly as proposed by Albrecht et al. J. Applied Physics, 1991, or indirectly as was done originally by Martin et al.
The indirect method uses a high Q cantilever, such that damping is small. The amplitude versus frequency curve of a high Q lever is very steep around the resonant frequency. Martin et al oscillated the lever near the resonant frequency and brought the tip close to the surface. The Van der Waals interaction with the surface shifts the resonance curve. This has the effect of shifting the resonance closer or further to the frequency at which the lever is oscillated, depending on which side of resonance the oscillation is at. Thus, indirectly, the amplitude of oscillation will either increase or decrease as a consequence of the resonance shift. The amplitude change is measurable (AM type detection). This change in amplitude close to the surface compared to the amplitude far away from the surface (the free amplitude) can be used as a setpoint to allow surface tracking. The direct method measures the frequency shift itself (FM type detection). Both methods are bound by the same interaction constraints.
FIG. 5 illustrates this non-contact operation. The tip is driven at a known amplitude and frequency of oscillation, which is typically near a cantilever resonance. The amplitude of this oscillation is detected by a deflection detector, which can be of various types described in the references. When the tip is sufficiently far away from the surface, it will oscillate at the free amplitudes A.sub.o, as shown in FIG. 5. As shown in FIG. 5, when the tip is brought closer to the surface, the Van der Waals interaction will shift the resonant oscillatory frequency slightly. This shift causes either an increased or decreased amplitude. A.sub.s, or the frequency shift may be measured directly. This modified amplitude value may be used as a setpoint in the manner of other above described SPM's, such that as the tip is scanned laterally, the tip height may be adjusted with feedback to keep setpoint, A.sub.s, at a constant value. Thus an image of the surface may be generated without surface contact, and without electrical interaction as needed by a scanning tunnelling microscope STM. The resonant shift may also be caused by other force interactions, such as magnetic field interaction with a magnetic tip. Thus this type of AFM may in theory be easily configured to map a variety of parameters using the same or similar construction.
The Van der Waals force is very weak, and decreases rapidly with separation, so the practical furthest distance for measurable interaction is 10 nm above the surface, as shown in FIG. 1, taken from Sarid, Scanning Force Microscopy, Oxford University Press, 1991. To shift the resonance of the lever, the lever must oscillate within this envelope of measurable force gradient. If just a small portion of the oscillation is within the envelope, the resonance will not be appreciably affected. Thus the oscillation amplitude must be small. A compendium of all non-contact AFM research can be found in Scanning Force Microscopy by Sarid, above noted, no researcher was able to operate a non-contact AFM with a free oscillation amplitude of greater than 10 nm. This limitation as will be shown limits the usefulness of the non-contact method.
Although developed at essentially the same time as the contact AFM, the non-contact AFM has rarely been used outside the research environment due to problems associated with the above constraints. The tip must be operated with low oscillation amplitude very near the surface. These operating conditions make the possibility very likely of the tip becoming trapped in the surface fluid layer described by Hansma et al. This effect is illustrated in FIG. 6, an amplitude versus displacement curve. A cantilever with probe is oscillated at a free amplitude A.sub.o, and the vertical position of the fixed end of the lever is varied from a height where the probe is not affected by the surface to a point where the probe is captured by the surface and oscillation ceases. The curve is typical for oscillation amplitudes of 10 nm or less. Such curves have been measured by the inventors, and were also described by Martin et al, and also by Ducker et al, in "Force Measurement Using an AC Atomic Force Microscope", J. of Applied Physics, 67(9), 1 May 1990. As the curve clearly shows, when the tip is brought near the surface there is a narrow region where the amplitude is affected by the Van der Waals interaction before it becomes abruptly captured by the surface fluid layer, and oscillation becomes very small It is this narrow region in which the non-contact AFM must operate As a surface is scanned, any variations in the surface topography may cause the tip to become captured if the feedback cannot perfectly respond to the topography variations. If the tip does become captured, the control system will lift the fixed end of the lever until the tip breaks free, and then re-establish the setpoint. As can be seen from FIG. 6, there is significant hysteresis in the withdraw process, which will cause serious instability in the image data. Thus non-contact microscopes must scan very slowly so the feedback loop has sufficient time to prevent the tip becoming stuck to the surface. Moreover, because the tip must be operated above the fluid layer, the lateral resolution is inferior to the contact mode. Typically, the noncontact AFM must operate with the tip 5-10 nm above the surface, which limits the lateral resolution to 5-10 nm. Contact mode AFM's typically have lateral resolution of better than 1 nm.
For measuring the frequency shift using amplitude detection, the sensitivity depends on the cantilever having a very sharp resonance peak, which in turn gives a very slow response time because undamped systems require a long time to recover from a perturbation. Thus, sensitivity and response time are inversely coupled. The high Q requirement also places restrictions on the design of the lever to minimize the effect of air as a damping agent. One could improve the time response by using cantilevers which may be operated at a higher frequency, but such levers are stiffer and therefore have reduced sensitivity to the Van der Waals interaction. Thus it can be seen that high sensitivity and fast response are very difficult to achieve with a non-contact AFM. Furthermore, the weak force interaction places restrictions on the height at which the tip may be operated and the amplitude of oscillation. The presence of the fluid layer near this height makes capture of a lever with a small oscillation likely, so slow time response is a serious stability problem. For these reasons, despite their many potential advantages, non-contact AFM's have yet to be successful commercially.
The non-contact AFM has been used successfully in the measurement of magnetic fields on objects such as magnetic storage media. With a tip of, or coated with, magnetic material, the force interaction between the tip and magnetic sample is much stronger than the Van der Waals interactions, and is longer range. Thus, the non-contact FM (also called magnetic force microscope, MFM) may be operated without the need for ultra-high sensitivity, as required for surface profiling. However since magnetic fields are seldom continuous, some interaction is necessary to guide the tip over the surface between magnetic regions. Rugar et al, (Magnetic Force Microscopy, IBM Research Report, Almaden Research Center, Dec. 12, 1990) found that applying an electric field between the tip and sample would produce a larger effect than the Van der Waals force, so the hard disks could be scanned without the probe sticking to the surface. This method limits the technique to conductive surfaces.